Nvidia invests $3.5B in MediaTek AI deal

Nvidia is investing $3.5 billion in MediaTek through convertible bonds. The two companies announced a major partnership expansion this week, with MediaTek adopting Nvidia's NVLink Fusion platform for its custom AI accelerators, local AI systems, and automotive platforms.

This move allows Nvidia to secure a foothold in the growing custom AI accelerator market. While Nvidia dominates the merchant AI GPU sector, major tech firms are increasingly developing their own bespoke chips. Companies like AWS, Google, Meta, Microsoft, and OpenAI are pursuing custom solutions partly to reduce reliance on expensive off-the-shelf GPUs.

Nvidia cannot stop this trend. Instead, the NVLink Fusion strategy ensures Nvidia's technology remains central even when its GPUs are not. If a customer replaces some Nvidia hardware with their own custom XPUs, Nvidia wants those XPUs connected via NVLink and deployed within its networking and rack architectures.

The NVLink Fusion Platform

MediaTek will use NVLink Fusion as the foundation for custom accelerators designed to evolve with future Nvidia architectures. The platform consists of several key components.

This setup lets potential MediaTek customers focus on their unique compute architecture. Nvidia and MediaTek provide the surrounding connectivity, memory, packaging, manufacturing, and rack-level technologies. Essentially, customers get a pre-developed rack-scale platform for their custom AI accelerators.

Hyperscalers can develop their own rack-scale solutions. Smaller companies often lack the resources. This partnership offers them a turnkey alternative.

Strategy and Market Position

Nvidia tends to supply complete AI infrastructure platforms, not just accelerators. The deal with MediaTek fits this strategy perfectly. It spreads the NVLink Fusion platform beyond Nvidia's own products, popularizing it among users of custom hardware.

Nvidia founder and CEO Jensen Huang commented on the alliance. "MediaTek is one of the world's great semiconductor companies, with exceptional expertise in system-on-chip design, connectivity, leading performance and power efficiency," Huang said. "Together, we are building platforms that bring Nvidia accelerated computing to new markets."

The collaboration aims to give customers freedom to create differentiated AI systems at enormous scale.

Expanding Collaboration Areas

The companies are also broadening their work on local AI computing. They previously collaborated on the GB10 Grace Blackwell Superchip for the DGX Spark system. That partnership was considered positive.

Now, Nvidia and MediaTek plan to cooperate on multiple generations of RTX Spark and DGX Spark processors. These are intended for client systems, AI developer supercomputers, and enterprise workstations.

Finally, the multi-generation automotive collaboration will continue. MediaTek's Dimensity Auto platforms already integrate Nvidia AI and RTX graphics for intelligent vehicle cockpits, operating alongside Nvidia Drive AGX. Future generations will further combine MediaTek's automotive SoC expertise with Nvidia's accelerated computing, AI, graphics, and software to build more advanced software-defined and AI-powered vehicles.

According to the source report from Tom's Hardware, the investment is structured as convertible bonds issued by MediaTek.